JP6446993B2 - Voice control device and program - Google Patents

Description

The present invention relates to a voice control device and a program.

  In recent years, the following have been proposed as speech synthesis techniques. That is, by synthesizing and outputting speech corresponding to the user's speech tone and voice quality, a technique for sounding more humanly (see, for example, Patent Document 1), analyzing the user's speech, A technique for diagnosing a health condition or the like (see, for example, Patent Document 2) has been proposed. In addition, a voice dialogue system that recognizes a voice input by a user and outputs a content specified in a scenario by voice synthesis to realize a voice dialogue with the user has been proposed (for example, Patent Document 3). reference).

JP 2003-271194 A Japanese Patent No. 4495907 Japanese Patent No. 4832097

By the way, it is assumed that a dialogue system that combines the above-described voice synthesis technology and a voice dialogue system and retrieves data and outputs it by voice synthesis in response to a user's voice question. In this case, a problem has been pointed out that the voice output by the voice synthesis feels unnatural to the user, specifically, the machine sometimes feels roaring.
The present invention has been made in view of such circumstances, and one of its purposes is to provide a voice control device and a program capable of giving a user a natural feeling.

  First of all, when considering a man-machine system that outputs the answers to questions from users by speech synthesis, first of all, what kind of dialogue is made between people, other than linguistic information, especially dialogue. Considering the pitch (frequency) to be characterized.

  As a dialogue between people, a case where the other person (b) responds to a question (question) by one person (a) will be considered. In this case, when a asks a question, not only a but also b trying to answer the question often leaves a strong impression of the pitch in the specific section of the question. b, when responding with consent, approval, affirmation, etc., the pitch of the part that characterizes the answer has a specific relationship, specifically a Kyowa interval, Speak to be in a relationship. A who has heard the answer has a relationship between the pitch that remains in the impression about his question and the pitch of the part that characterizes the answer to the question. It is thought that you will have a good impression.

  In this way, in the dialogue between people, it can be considered that the pitch of the question and the pitch of the answer are not irrelevant and have the above relationship. Based on these considerations, when considering a dialogue system that outputs (answers) answers to questions from users by speech synthesis, the following configuration has been adopted in order to achieve the above objective for the speech synthesis. .

  That is, in order to achieve the above object, a speech synthesizer according to an aspect of the present invention includes a first pitch acquisition unit that acquires a pitch of a specific section among questions based on an input voice signal, and the question. An answer acquisition unit for acquiring the voice data of the answer to the voice, a second pitch acquisition unit for acquiring the pitch based on the voice data of the acquired answer, and a predetermined sound for the pitch based on the voice data of the answer A pitch shift amount determination unit that determines a pitch shift amount to a target pitch that is a target pitch within a high range and maintains a specific relationship with respect to the pitch of the specific section; and the answer An answer composition unit for synthesizing an answer by shifting the pitch based on the voice data by the pitch shift amount.

According to this embodiment, according to the present embodiment, the answers to the questions issued by the user can be synthesized (reproduced) without being unnatural and preventing deterioration in the quality of hearing.
Note that the answer is not limited to a specific answer to the question, but includes an answer (interjection). In addition to human voices, answers include animal calls such as “bow” and “meow”. That is, the answer and the voice here are concepts including not only a voice uttered by a person but also an animal cry.
Among the questions, the pitch of a specific section refers to the pitch of the part left with a strong impression. Specifically, the peak value of the section where the volume is equal to or higher than the predetermined value, or the end section of the question It is preferable that the pitch is.
The pitch based on the voice data is, for example, the pitch at the characteristic part when the voice data is played back as a standard, and the characteristic part is the part with the highest pitch and volume at the beginning of the word. As well as the average pitch.
Here, the specific relationship is preferably a Kyowa interval relationship. Kyowa means a relationship in which a plurality of musical sounds are generated at the same time and are well-harmonized with each other. These pitch relationships are called Kyowa pitches. The degree of cooperation is higher as the frequency ratio (frequency ratio) between two sounds is simpler.

  In the above aspect, the pitch shift amount determination unit may change the pitch shift amount in octave units so that the pitch based on the voice data of the answer falls within the pitch range. . When the audio data is shifted by the pitch shift amount, if the shift amount is large, the sound data is deteriorated. However, according to the aspect, such deterioration can be prevented.

The first pitch acquisition unit preferably acquires the highest pitch value of a section in which the volume of the input audio signal is equal to or higher than a predetermined value as the pitch of the specific section. In this determination, the determination may be made with a hysteresis characteristic or may be weighted on the condition that the pitch can be detected.
The aspect of the present invention can be conceptualized as a program that causes a computer to function as the speech synthesizer as well as the speech synthesizer.

It is a block diagram which shows the structure of the speech synthesizer which concerns on embodiment. It is a flowchart which shows operation | movement of a speech synthesizer. It is a flowchart which shows operation | movement of a speech synthesizer. It is a figure which shows the example of a pitch of the question by a user, and the reply by a speech synthesizer. It is a figure for demonstrating the premise of an application example. It is a figure which shows the principal part of the process in an application example (the 1). It is a figure which shows the principal part of the process in an application example (the 2). It is a figure which shows the principal part of the process in an application example (the 3). It is a figure which shows the operation | movement outline | summary of an application example (the 4).

  Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a diagram showing a configuration of a speech synthesizer 10 according to an embodiment of the present invention.
The speech synthesizer 10 is a device that is incorporated in a stuffed animal, for example, and synthesizes and outputs an answer such as a question when a user asks the stuffed animal. The voice synthesizer 10 includes a CPU (Central Processing Unit), a voice input unit 102, and a speaker 142. When the CPU executes an application program installed in advance, a plurality of functional blocks are as follows. Built. Specifically, in the speech synthesizer 10, a speech feature amount acquisition unit 106, an answer selection unit 110, an answer pitch acquisition unit 112, a pitch shift amount determination unit 114, and an answer synthesis unit 116 are constructed.

  Although not particularly illustrated, the speech synthesizer 10 also includes a display unit, an operation input unit, and the like, so that the user can check the status of the device and input various operations to the device. Various settings can be made. The voice synthesizer 10 is not limited to a toy such as a stuffed toy, and may be a so-called pet robot, a terminal device such as a mobile phone, a tablet personal computer, or the like.

  Although not described in detail, the sound input unit 102 includes a microphone that converts sound into an electrical signal and an A / D converter that converts the converted sound signal into a digital signal.

The voice feature amount acquisition unit 106 (first pitch acquisition unit) analyzes the voice signal converted into the digital signal, and classifies the voice signal into a speech segment and a non-speech segment, and among the speech segments, The pitch of a specific section in the voiced section is detected, and data indicating the pitch is supplied to the answer selection unit 110 and the pitch shift amount determination unit 114. Here, the utterance section refers to, for example, a section in which the volume of the audio signal is equal to or higher than a threshold, and conversely, the non-utterance section refers to a section in which the volume of the audio signal is less than the threshold. The voiced section is a section in which the pitch (pitch) of the voice signal can be detected in the utterance section. The section in which the pitch can be detected means that there is a periodic part in the audio signal and that part can be detected.
Here, the specific section is the end section of the voiced section, and the pitch is the highest value in the end section. Further, the end section is a section of a predetermined time (for example, 180 msec) that is temporally forward from the end of the voiced section. As will be described later for the voiced section, the volume of the voice signal may be determined by two (or three or more) threshold values.

  The answer library 124 stores a plurality of voice data of answers to questions by the user in advance. This voice data is a recording of the voice of a model person. For example, “Yes”, “No”, “Yes”, “Yes”, “Fun”, “I see” For example, it is a companion. The voice data of the answer is in a format such as wav or mp3, for example, and a pitch for each waveform sample (or every waveform cycle) when reproduced as a standard and an average pitch obtained by averaging them are obtained in advance. Data indicating the average pitch (pitch based on the answer) is stored in the answer library 124 in association with the voice data. The standard reproduction here means that the audio data is reproduced under the same conditions as the recording conditions (sampling frequency).

When the data indicating the pitch of the specific section is output from the voice feature quantity acquisition unit 106, the answer selection unit 110 (answer acquisition unit) selects one voice data of the answer to the voice from the answer library 124. The voice data of the selected answer is read and output together with the data indicating the associated average pitch.
For example, the answer selection unit 110 may select a single piece of voice data from among a plurality of pieces of voice data. For example, the answer selection unit 110 may be random. It is also possible to select near voice data.

In the present embodiment, the meaning content of the user's question is not taken into account for the selected answer. However, the speech synthesizer 10 is used as a response to the question issued by the user. This is sufficient for a returning device.
On the other hand, a language analysis unit 108 is provided as shown by a broken line in the figure, the language analysis unit 108 analyzes the meaning content of the question specified by the audio signal, and the answer selection unit 110 transmits the database via a database or the like. It is good also as a structure which produces the answer with respect to the said question.

The answer pitch acquisition unit 112 (second pitch acquisition unit) extracts data indicating the average pitch of answers from the data read by the answer selection unit 110 and supplies the extracted data to the pitch shift amount determination unit 114. To do.
The pitch shift amount determination unit 114 determines the answer based on the difference between the pitch of the specific section in the voice signal output from the voice feature value acquisition unit 106 and the average pitch of the answers output from the answer pitch acquisition unit 112. The pitch shift amount when reproducing the audio data is determined as will be described later.
The answer synthesizing unit 116 reproduces (synthesizes) the answer voice data read from the answer library 124 by shifting it by the pitch shift amount determined by the pitch shift amount determining unit 114. Note that the audio signal whose pitch is shifted is converted into an analog signal by a D / A converter (not shown), and then acoustically converted by the speaker 142 and output.
The data associated with the pitch of the answer is stored in the answer library 124, and the data used for determining the pitch shift amount by the pitch shift amount determining unit 114 is the average pitch. It may be other than data indicating. For example, an intermediate value of the pitch may be used, or an average pitch of a predetermined section of the audio data may be used.

Next, the operation of the speech synthesizer 10 will be described.
FIG. 2 is a flowchart showing processing operations in the speech synthesizer 10.
First, when the user asks a question about the stuffed animal to which the speech synthesizer 10 is applied, the processing shown in this flowchart is started. Here, for convenience, the case where the pitch of the answer voice data is higher than the pitch of the voice (question) of the user will be described as an example.

First, in step Sa <b> 11, the audio signal converted by the audio input unit 102 is supplied to the audio feature amount acquisition unit 106.
Next, in step Sa12, the voice feature quantity acquisition unit 106 performs an analysis process on the voice signal from the voice input unit 102, that is, a process of detecting the pitch of a question issued by the user.
In step Sa13, the answer composition unit 116 determines whether or not the answer is being reproduced.

If the answer is not being played back (if the determination result in step Sa13 is “No”), the voice feature acquisition unit 106 determines whether or not the question (speech) of the voice signal from the voice input unit 102 has ended. A determination is made (step Sa14). Note that whether or not the inquiry has ended is specifically determined based on, for example, whether or not a state in which the volume of the audio signal has become less than a predetermined threshold has continued for a predetermined time.
If the inquiry has not ended (if the determination result in step Sa14 is “No”), the processing procedure returns to step Sa11, whereby the audio feature quantity acquisition unit 106 receives the audio signal from the audio input unit 102. Continue the analysis process.
If the question has been completed (if the determination result in step Sa14 is “Yes”), the pitch shift amount determination unit 114 reproduces the pitch when the voice data of the answer selected by the answer selection unit 110 is reproduced. The shift amount is determined as will be described later (step Sa15).
Then, the pitch shift amount determination unit 114 notifies the answer composition unit 116 of the determined pitch shift amount, and instructs the reproduction of the voice data of the response selected by the response selection unit 110 (step Sa16). In accordance with this instruction, the answer synthesizer 116 shifts and reproduces the audio data by the pitch shift amount determined by the pitch shift amount determiner 114 (step Sa17).

  In step Sa13, when the answer composition unit 116 determines that the answer is being reproduced (when the determination result in step Sa13 is “Yes”), the answer is being reproduced according to a certain question. For example, when the following question is asked by the user. In this case, the processing procedure does not return the path of steps Sa14 and Sa11, and proceeds to step Sa17, so that the reproduction of the answer has priority.

FIG. 3 is a flowchart showing details of the process of step Sa15 in FIG. 2, that is, the process of determining the pitch shift amount of the answer voice data.
The premise for executing this process is that the answer composition unit 116 is not reproducing the answer (the determination result in step Sa13 is “No”), and the user has finished inputting the question ( The determination result of step Sa14 is “Yes”).
First, in step Sb11, the pitch shift amount determination unit 114 acquires data indicating the pitch of the specific section in question from the voice feature amount acquisition unit 106.

  On the other hand, the answer selection unit 110 selects voice data of the answer to the question by the user from the answer library 124, the voice data of the selected answer, and data indicating the average pitch associated with the voice data Is read. Among these, the answer pitch acquisition unit 112 supplies data indicating the average pitch among the read data to the pitch shift amount determination unit 114. Thereby, the pitch shift amount determination unit 114 acquires data indicating the average pitch of the answer selected by the answer selection unit 110 (step Sb12).

  Next, the pitch shift amount determination unit 114 temporarily determines a pitch having a predetermined relationship (for example, 5 degrees below) with respect to the pitch of the specific section in question as a pitch when replying with voice data. (Step Sb13).

  Subsequently, the pitch shift amount determination unit 114 determines the pitch temporarily determined from the average pitch of the answers selected by the answer selection unit 110 (the pitches changed in steps Sb16 and Sb18 described below in addition to step Sb13). Is calculated (step Sb14). The pitch shift amount determination unit 114 determines whether or not the pitch (pitch after the shift) when the average pitch of the answer is shifted by the pitch shift amount is lower than the lower limit threshold (step Sb15). Here, the lower limit threshold value is a threshold value indicating how much a lower pitch is allowed with respect to the average pitch of answers, and will be described in detail later.

  If the pitch after the shift is lower than the lower limit threshold (if the determination result in step Sb15 is “Yes”), the pitch shift amount determination unit 114 raises the pitch of the tentatively determined answer by one octave, and The pitch raised by one octave is provisionally determined again as the pitch when replying with voice data (step Sb16). After this, the processing procedure returns to step Sb14, the pitch shift amount is calculated again, and the determinations of steps Sb15 and Sb17 are executed.

On the other hand, if the pitch after the shift is not lower than the lower limit threshold (if the determination result in step Sb15 is “No”), the pitch shift amount determination unit 114 determines that the pitch after the shift is lower than the upper limit threshold. It is determined whether or not it is high (step Sb17). Here, the upper limit threshold value is a threshold value indicating how much pitch is allowed to be higher than the average pitch of answers, and details will be described later.
If the pitch after the shift is higher than the upper limit threshold value (if the determination result in step Sb17 is “Yes”), the pitch shift amount determination unit 114 lowers the pitch of the tentatively determined answer by one octave, and The pitch lowered by one octave is provisionally determined again as the pitch when answering with voice data (step Sb18). After this, the processing procedure returns to step Sb14, the pitch shift amount is calculated again, and the determinations of steps Sb15 and Sb17 are executed.

  If the pitch after the shift is not higher than the upper limit threshold (if the determination result of step Sb17 is “No”), the pitch after the shift is a predetermined pitch that is not less than the lower limit threshold and not more than the upper limit threshold. Means it is within range. For this reason, the pitch shift amount determination unit 114 shifts the processing procedure to step Sb19, determines the pitch that is currently in the tentative determination stage as the final determination, and notifies the answer composition unit 116 of the pitch shift amount.

FIG. 4 is a diagram illustrating the relationship between a question input by a user and an answer synthesized by the speech synthesizer 10 with the pitch as the vertical axis and the time as the horizontal axis.
In this figure, the solid line indicated by the symbol T1 simply indicates a change in the pitch of the question by the user as a straight line. A symbol P1 is a pitch of a specific section in the question T1.
Further, in the figure, the solid line indicated by reference symbol A1 is a diagram simply showing the change in pitch when the voice data of the answer selected for the question T1 is reproduced as a standard, and reference symbol P2 indicates the average It is pitch.

When the pitch of the answer A1 is reproduced without shifting to the question T1, it is easy to receive a mechanical feeling. For this reason, in the present embodiment, firstly, the pitch is in a relationship that is, for example, 5 degrees below the Kyowa interval with respect to the pitch P1 of the specific section (ending) that is a characteristic and impressive part of the question T1. An attempt is made to reproduce the answer A1-1 by shifting the answer A1 so as to be P2-1. Note that the symbol D1 is a pitch difference between the pitch P1 and the pitch P2-1.
However, if the pitch shift amount D2 of the answer A1-1 with respect to the answer A1 is too large, the auditory quality deteriorates when the answer A1-1 that has been pitch-shifted is reproduced. In particular, if the pitch of the specific section of the question is far away from the average pitch of the answer (for example, if the user making the question is male and the answer model is female), the direction of decreasing the pitch If the playback is shifted to, it tends to be unnatural and remarkably deteriorated.

Therefore, in the present embodiment, second, while maintaining that the pitch of the answer synthesized by the answer synthesis unit 116 has a specific relationship with the pitch P1, the average pitch P2 of the original answer A1 is maintained. On the other hand, the pitch P2-1 of the answer A1-1 is shifted step by step in units of octaves until it falls within a predetermined pitch range. In the example of FIG. 4, the answer A1-4 is increased by 3 octaves from the answer A1-1 to the answer A1-2 and the answer A1-3 until it falls within the pitch range based on the pitch P2 of the answer A1. It is an example.
In FIG. 4, the pitch difference amount from the average pitch P2 to the lower limit threshold Pth_L among the pitch ranges set with reference to the average pitch P2 of the answer A1 is defined by the code T_L, and the upper limit threshold Pth_H. The pitch difference amount until is defined by the code T_H. That is, the lower limit threshold Pth_L is a relative value defined by the pitch difference amount T_L based on the average pitch P2 of the answer A1, and similarly, the upper limit threshold Pth_H is a pitch based on the average pitch P2. It is a relative value defined by the difference amount T_H. Since there are a plurality of voice data of answers stored in the answer library 124, the lower limit threshold value Pth_L and the upper limit threshold value Pth_H that define the pitch range of the answer are different for each answer. By relatively defining the pitch difference amount with reference to P2, there is no need to store the lower limit threshold value Pth_L and the upper limit threshold value Pth_H in advance for each answering voice data.
Note that the pitch P2-1 is in the relationship of the concert pitch with respect to the pitch P1 of the question T1, and the pitch P2-4 is in a relationship of 3 octaves above the pitch P2-1. For this reason, since the relationship of the integer ratio is maintained between the frequency of the pitch P2-4 and the frequency of the pitch P2-1, the pitch P1 and the pitch P2-4 are almost the same. The relationship of Kyowa intervals will be maintained.

  In addition, for example, the answer A1-2 may be determined and reproduced, but if only one octave shift is performed from the answer A1-1, the shift amount from the original answer A1 is large and still unnatural. Or the degree of quality degradation in the sense of hearing may not be overlooked, so that it falls within a predetermined pitch range.

  According to this embodiment, an answer to a question issued by a user can be synthesized (reproduced) without being mechanical, unnatural, and preventing deterioration in audible quality. Further, it is not necessary to add attribute information indicating whether the answer is female or male to the voice data of the answer, and determine the pitch shift amount according to the attribute information.

In the present embodiment, as illustrated below, by combining the answers, it is possible to synthesize answers with emotions such as an angry answer and a careless answer.
FIG. 5 is a diagram for explaining the following terms. In the figure, the symbol Av is the pitch change width of the answer A1, and the symbol d is the answer A1 starting to play from the end of the question T1. The code Ad is the playback time of the answer A1. Moreover, the code | symbol Tg shows the time change of the sound volume in the question T1, and the code | symbol Ag shows the time change of the sound volume in the answer A1.

  For example, in the application example (part 1) shown in FIG. 6, the playback speed of the answer A1 is increased to play back like the answer A11, and the time d11 from the end of the question T1 to the start of playback of the answer A11 is expressed as time It is shorter than d, and the volume Ag11 of the answer A11 is larger than the volume Ag. As a result, an answer expressing anger can be output. Note that since the playback speed of the answer A11 is increased, the playback time Ad11 of the answer A11 is shorter than the playback time Ad of the answer A1.

  Further, for example, in the application example (part 2) shown in FIG. 7, the playback speed of the answer A1 is slowed down and played back as the answer A12, and the time d12 from the end of the question T1 to the start of playback of the answer A12 is set. , Longer than the time d, and the volume Ag12 of the answer A12 is made smaller than the volume Ag. This makes it possible to output a so-called unspoken answer. Since the playback speed of the answer A12 is slow, the playback time Ad12 of the answer A12 is longer than the playback time Ad of the answer A1.

  In addition, in the application example (part 3) shown in FIG. 8, the answer A13 is reproduced like the answer A13 so that the pitch increases toward the end with respect to the answer A1, that is, the answer A13 changes in pitch. Reproduction is performed so as to increase by the width Av13. This makes it possible to output an answer that asks questions.

Thus, when composing an answer with emotion, the pitch change width (including the height direction) of the answer to the question T1, the time from the end of the question T1 until the answer starts to be reproduced, the answer playback volume, The response playback speed or the like may be set by the user or the like via the operation input unit.
Moreover, it is good also as a structure in which a user can select the kind of answer which makes an anger answer, a careless answer, and a question ask.

  Moreover, you may detect an utterance area, a voiced area, etc. from the audio | voice signal of the question uttered by the user as follows.

FIG. 9 is a diagram illustrating a relationship between detection of a speech segment, a non-speech segment, and a voiced segment and a sound volume threshold in an application example (No. 4).
In this figure, the temporal change in pitch is shown in (a) and the temporal change in volume is shown in (b) for the question that the user has made. Specifically, it is shown that the pitch and volume gradually increase and then start to decrease from the middle.

Here, the threshold value Thvg_H is applied when the pitch (pitch) can be detected from the audio signal, and is applied when the volume of the question is in the increasing direction, and the utterance period when the volume exceeds the threshold Thvg_H. And the start of a voiced interval is detected.
The threshold value Thvg_L is applied when the pitch can be detected from the audio signal, and is applied when the volume of the question is in the descending direction, and is detected as the end of the voiced section when the volume is less than the threshold Thvg_L. .
In utterance, even if the volume falls below the threshold Thvg_L, there is a shake back of the volume. Therefore, in the example of this figure, a lower threshold value Thhuvg that can detect the pitch from the questioned audio signal is prepared, and the questioned sound volume is in a descending direction, and after the sound volume is less than the threshold value Thvg_L, Furthermore, when it becomes less than the threshold Thuvg, it is detected that the utterance section is ended (the start of the non-speaking section).
Note that the threshold values Thvg_H, Thvg_L, and Thhuvg are as follows:
Thvg_H>Thvg_L> Thuvg
Are in a relationship.

The highest value of the pitch in the voiced section detected by the threshold values Thvg_H and Thvg_L may be detected as the pitch of the specific section in the question.
Moreover, if the voiced section detected in this way is a relatively short time, it is assumed that noise is picked up as an audio signal. For this reason, as a condition for detection as a voiced section, when a pitch can be detected from a voice signal and the volume of the question is in an increasing direction, a predetermined time or more has elapsed since the threshold Thvg_H or higher. It is good also as a requirement.
If the non-voiced (unvoiced) section is a relatively short time, it is assumed that the question has not ended. Therefore, the pitch can be detected from the voice signal as a condition for detection as an unvoiced section. However, when the volume of the question is in the descending direction, it may be a requirement that a predetermined time elapses after becoming less than the threshold Thvg_L.
Of course, when an unvoiced section is detected as a requirement that a predetermined time has passed since the volume has become less than the threshold Thvg_L after a voiced section where it has been detected that the predetermined time has passed since the volume has become equal to or greater than the threshold Thvg_H In addition, the highest value of the pitch in the previous voiced section may be detected as the pitch of the specific section in the question.

In addition, when the user asks that the last segment of the voiced segment is an unvoiced sound (in short, a sound that is not accompanied by vocal cord vibration during utterance), the pitch of the unvoiced sound portion is changed from the immediately preceding voiced sound portion. May be estimated.
The specific section of the question by the user is the last section of the voiced section, but it may be the beginning section, for example, and the user arbitrarily sets which part of the question to specify the pitch It is good also as a structure which can be performed.
Moreover, instead of using two of the volume and the pitch for detecting the voiced section, it may be detected using either one, and the user determines which one is used to detect the voiced section. You may choose.

As for the voice data of answers stored in the answer library 124, answers of the same content may be stored across a plurality of people such as persons A, B, C,. For the persons A, B, C,..., For example, celebrities, talents, singers, etc., voice data is stored in a library for each person.
In the case of creating a library in this way, the answer voice data may be stored in the answer library 124 via a medium such as a memory card, or the voice synthesizer 10 is provided with a network connection function to answer from a specific server. May be downloaded and stored in the answer library 124. When obtaining the answer voice data from the memory card or the server, it may be free or paid.
On the other hand, it can be configured so that the user can select the person who wants to answer the model as a model by the operation input unit, etc., or for each condition (day, week, month, etc.) It is good also as a structure determined at random.

As for the voice data of answers, a library of voices recorded by the user himself / herself, the user's family and acquaintances (or converted into data by a separate device) via the microphone of the voice input unit 102 is a library. May be used.
When the answer is made with the voice of a person close to the person like this, it is possible to obtain a feeling as if the person is interacting with the person when the question is made.

  In addition, the answer may be a call from an animal (dog, cat, etc.), or may be configured such that a dog breed or the like can be selected as appropriate. In this way, by using an answer as an animal call, it is possible to obtain a kind of healing effect as if it were interacting with the animal.

  The answer pitch acquisition unit 112 analyzes the voice data of the answer determined by the answer selection unit 110, acquires an average pitch when the voice data is reproduced as a standard, and uses the data indicating the pitch as a sound. A configuration may be adopted in which the high shift amount determination unit 114 is supplied. According to this configuration, it is not necessary to associate the data indicating the pitch with the voice data of the answer and store it in the answer library 124 in advance.

  In the embodiment, the case where the pitch of the voice data of the answer is higher than the pitch of the question asked by the user has been described as an example. This is also applicable when the pitch of the is low.

DESCRIPTION OF SYMBOLS 102 ... Voice input part, 106 ... Voice feature-value acquisition part (1st pitch acquisition part), 110 ... Answer selection part, 112 ... Answer pitch acquisition part (2nd pitch acquisition part), 114 ... Pitch shift amount Decision unit 116... Response composition unit 124.

Claims (10)

A first pitch acquisition unit for acquiring a pitch of a specific partial section among the questions based on the input voice signal;
An answer obtaining unit for obtaining an answer to the question;
A second pitch acquisition unit for acquiring a pitch based on the voice signal of the acquired answer;
Sound up to a target pitch that is within a predetermined pitch range with respect to the pitch based on the voice signal of the answer and that maintains a specific relationship with the pitch of the partial section A pitch shift amount determination unit for determining a high shift amount;
A voice control unit that shifts the pitch based on the voice signal of the answer by the pitch shift amount;
A voice control device comprising: The pitch shift amount determination unit
The voice control device according to claim 1, wherein the pitch shift amount is changed in octave units so that the pitch based on the voice signal of the answer falls within the pitch range. The first pitch acquisition unit
The voice control device according to claim 1 or 2, wherein a maximum pitch value of a section in which a volume of an input voice signal is equal to or higher than a predetermined value is acquired as a pitch of the partial section. The first pitch acquisition unit
The pitch of the last section of the voiced section of the input audio signal is acquired as the pitch of the partial section.
The voice control device according to claim 1, wherein The tail section is a section of a predetermined time from the end of the time when the volume of the audio signal capable of detecting the pitch decreases and falls below a predetermined threshold.
The voice control apparatus according to claim 4. The first pitch acquisition unit
The pitch of the beginning section of the input voice signal is acquired as the pitch of the partial section.
The voice control device according to claim 1, wherein The user can arbitrarily set which part of the input audio signal is the partial section.
The voice control device according to claim 1, wherein The partial section is a section of the input audio signal that leaves a strong impression on the person who has heard the audio signal.
The voice control device according to claim 1, wherein The pitch based on the voice signal of the answer is an average value or an intermediate value of the pitch of the voice signal of the answer.
The voice control device according to claim 1, wherein Computer
A first pitch acquisition unit for acquiring a pitch of a specific partial section among the questions based on the input voice signal;
An answer acquisition unit for acquiring an audio signal of an answer to the question;
A second pitch acquisition unit for acquiring a pitch based on the voice signal of the acquired answer;
Sound up to a target pitch that is within a predetermined pitch range with respect to the pitch based on the voice signal of the answer and that maintains a specific relationship with the pitch of the partial section A pitch shift amount determination unit for determining a high shift amount, and
A voice control unit that shifts the pitch based on the voice signal of the answer by the pitch shift amount;
A program characterized by functioning as

Images